TOP
Human electrocardiograms (ECG) are tracings that measure electrical activity generated by the heart. In a healthy heart, the electrical activity causes a coordinated contraction of the heart muscle. Many heart diseases have distinctive ECG patterns, and ECGs from healthy hearts display a wide variety of normal differences. Life insurance companies have long used ECGs as part of the underwriting assessment of Proposed Insureds (PI) for Individual Life and Disability policies.
In late 2022 Gen Re Medical Directors first noticed bizarre ECGs present in PI files from primary insurers. These “perfect” ECGs were performed by paramedical examiners and included with facultative files for reinsurance coverage. Initially they were very rare, but we have seen them become more common.
Recently three identical “perfect” ECGs on three different PIs were received from a single primary insurer, and we notified that company’s chief medical director. The paramedical company was notified and determined that the ECGs were indeed unrelated to any human test and were machine-made ECGs from the demo mode saved in their portable ECG machines.
The demo mode was apparently turned on at some point – and possibly can be turned on and off by the paramedical examiner. Portable ECG machines are used by parameds to take ECGs in the homes of PIs. They don’t require a connection by the 10 wire leads to the PI, so the image is saved in the machine as a demo. The demo ECG can be printed out and signed by paramed examiner and the PI just like a real ECG, so the PI could have no idea.
Demo ECGs look perfect: a perfectly flat baseline, perfectly symmetrical P and T waves, and have inverted waves in aVR. An example is below. However, tracings of any biologic system, including human ECGs, rarely achieve such precision and symmetry – and all three of the PI ECGs were identical, indicating a computer-based source.
This was not a product of artificial intelligence.
Insurance underwriters need to be aware they may receive significant computer-generated fake data. The causes of the fake data could be accidental, incompetence or fraud. Insurers must rely on the sources of the information they receive, in this case a paramedical exam company, to be as diligent as their company’s own underwriters are regarding authenticity of information.
